What’s Keeping a Prominent Election Analyst Up at Night

With the hours ticking down to Election Day, nervous political obsessives are watching every single poll, attempting to glean some new information that goes beyond “The race between Kamala Harris and Donald Trump is very close.” But what if those polls are off again? This possibility concerns Sean Trende, the senior elections analyst at RealClearPolitics, who is one of the most astute observers of American elections. (Among other things, he was early to warn of Democrats’ big problems with white working-class voters and to observe that Jon Ossoff and Raphael Warnock had a real chance to win their Senate races in Georgia.) I spoke with him about where he sees the race heading, the importance of abortion referendums to the outcome, and what makes him nervous about polling after two significant misses in a row.

On October 23, you wrote an article with the headline “It’s Close, But the Signs Aren’t Good for Harris.” I’m wondering if that’s still your basic view of the election a few days out. 
I think the headline was a little bit overstated, in retrospect. I think, overall, the polls are close — all point to a very close race. But if you look at the bigger picture and everything else surrounding the race — I think to the extent we’re going to rely on vibes, they’re better for Trump. I think Trump’s had a couple rough news cycles since then, but I still think, on balance, once you get past the polls — which, again, point to a very close race — the way I might put it is I’d rather be playing his hand than hers.

And why is that?
Well, you look at what the campaigns are doing beyond what the polls say. Harris went to Texas, which I thought was perhaps an encouraging sign for her campaign, but he’s doing things like going to Coachella …

And you mentioned the Madison Square Garden rally in your piece. Of course, that didn’t end up going so well for Trump.
No. But these are the types of things you do when you think things are going well and you’re trying to maybe win the popular vote or save some House seats.

The counterpoint to that, though, is that Trump projects crazy confidence at all times, and his campaign does the same.
And that’s fair. I had the line, which with Trump is only a half-joke, that maybe he just always wanted to play the Garden.

The way it played out does give me some pause, or some additional pause, I guess. But I still think I’d rather have his hand than hers. You look at what’s going on with early voting in Nevada, which is hard to read stuff into. But Jon Ralston, who’s accidentally forgotten more about early voting in Nevada than I have ever known, is pretty bearish on Democrats. And he is not that bearish that often.

It’s a commonly held view among analysts that his analysis on this is the only one that matters. 
Part of it is that Jon’s just really good at it and he’s done it a long time and he’s an honest broker, but part of it is that there isn’t much of an Election Day vote in Nevada. I think in 2020, 10 percent of the vote was cast on Election Day, and in 2022, 20 percent was. So when you’re looking at the early-voting data, you’re getting a really good sense in that state of what Election Day is going to look like or what the overall numbers are going to look like. And they’re really bad for Democrats. Now Republicans have changed their strategy. Donald Trump’s not telling his supporters not to vote early; he’s telling them to vote early, which is a smart move.

A masterstroke.
For Trump, I think it counts as one, but those numbers are scary. And to the extent you can look at other states, it really does look like minority turnout is down.

You’re looking at Georgia and North Carolina there?
Those are the two states I was going to say. The thing that got me about North Carolina is that Republican and independent votes are about what they were, but Democratic votes are down from where they were. And, again, you don’t know. Maybe people are going to shift and vote on Election Day; maybe they’re going to vote later in the early-voting cycle. It really is entrail-reading.

It’s almost like choosing your own adventure. I see people saying, “Well, the gender split is very advantageous for Democrats — more women than men.” And I have no idea whether that means anything. 
If I’m Republicans, I would much rather have a bunch of Republican women voting than a bunch of Democratic men. But yeah, the early-voting stuff really is entrail-reading. You can think of the election as an equation — like, a party’s vote share is the number of early votes times its share of the early votes plus the number of Election Day votes times its share of Election Day votes. And, really, all we know are the number of early votes.

I was told there would be no math.
Well, maybe you can play that part back a couple times. We know how many votes have been cast early to date and then people try to guess what the partisan breakdown is, even though we don’t really know. And then we know nothing about the Election Day vote. So unless you’re in a state like Nevada, where the Election Day vote is just tiny, it is really tough to read things into it.

You also wrote recently that the abortion referendums in Arizona and Nevada were being a bit underplayed, and that made you more bullish on Harris in those states. Do you still think that that could provide some kind of boost for her? 
It’s not just a turnout thing, although it can be. But if you’re a moderate pro-choice Republican woman and you go in and you think maybe you’re voting Trump-Vance because you think Harris is going to raise your taxes or whatever and then you look at your ballot and there’s an abortion question on it, it primes your thought process. So for wavering voters, I think that priming can have an effect. How much it’s worth, I don’t know, but I could see it being worth a couple points of people who go in kind of wishy-washy, but maybe they’re leaning Trump fans — and then they see the abortion referendum and they decide they’re going Harris. It’s like having one of her core messages right there on the ballot.

And those questions have done very well for Democrats pretty much everywhere they’ve been, often way surpassing the vote share of their candidates.
I think that’s right. Michigan is one of the states that had a big pro-Republican skew in the polls in 2022, and I honestly do think it’s the abortion referendum there. So, yeah, that could easily move things a couple points.

The simplest path for Harris still seems to be the Midwest “Blue Wall” states — Pennsylvania and Michigan, Wisconsin — which, on the whole, are a bit stronger for her than all the other states we’ve mentioned so far. Do you agree with that assessment?
I think that’s right. If Trump doesn’t lose any states from 2020, which I don’t think most people think he will — except maybe North Carolina — and adds Georgia, Arizona, and Nevada, he gets to 268. So he really needs one of those three: Wisconsin, Pennsylvania, or Michigan. And Harris’s polling has held up pretty well there. That was the story before Biden had his debate. That was the story for him, too, and that’s basically what we’ve gotten back to. The Sun Belt states seem to be slipping away. Trump’s doing well in the popular vote, but those Rust Belt states are holding tight. But then I start thinking to myself that an ABC News–Washington Post poll had Biden up by 17 in Wisconsin.

Ah, the infamous 17. 
Yeah. Those are the states where the polls were terrible.

I’ve talked to Charles Franklin, who runs the Marquette poll in Wisconsin, and others about how they’ve tried to fix the state errors. Nobody really knows whether they’ve successfully done it. 
Again, when you’re asking, “What’s your vibes reaction to the election?,” I do think you have to give some weight to the possibility that the pollsters are going to underestimate him. Maybe not by 17 points like WaPo did. But even if they’ve only fixed half of the problems from 2020 …

And then, of course, there’s a possibility that they’ve overcompensated, which many have pointed out as a possibility.
The thing that I come back to when I think about Trump voters who turn out in presidential elections is — when I go through back roads here in Ohio, there’s the house where you always wonder, Gosh, who lives there?, and the front yard is just completely decorated in Trump signs. The theory is that maybe Trump supporters, all other things being equal, just have low levels of social trust — and you can’t weight for that. Maybe the low level of social trust is why they don’t talk to pollsters.

That’s the most compelling theory I’ve heard.
There’s just nothing you can do about it. And that keeps me awake at night.

There’s also been some talk of pollster “herding” recently, the idea that all these surveys come out and they show nearly exactly the same thing. Do you see any evidence that that’s happening? Are people worried that they’re going to get it so wrong again that they’re not revealing their true numbers?
There’s some of that. It’s a one-way attitude, because the biggest penalty of the four outcomes is saying, “Trump’s going to win,” and he loses. If you say Harris is going to win and she loses, you’ve told a lot of people in media circles what they wanted to hear, and so they’re mad you’re wrong. But if you say Trump’s going to win and he loses, you’re a laughingstock.

I don’t know if I agree. If you say Harris is going to win and she loses, people are going to be devastated. They’re going to blame you. But if Harris wins, media circles will be in a good mood, so they won’t really care.
Natalie Jackson had a good rejoinder on herding, which is that what’s going on with polling these days is that, because response rates are so low, everything’s modeled and weighted. And when you do that, you lose some of the natural variance. That causes its own problems, because any time you model or weight, you’re injecting the pollsters’ views of things. It’s not like the old days where you would take your sample, maybe weight for race, and then send it out into the world — where you would get that kind of wide variance that you expect. All the data is so massaged that it just becomes more stable. So it’s not necessarily herding, but it’s things that can have the same effect. And that does make me nervous. We should have more outliers than we have.

We had two examples that came out from 2014 where pollsters showed the Virginia Senate race close, and they just didn’t publish the result.

That’s pretty egregious.
I think it’s the same thing: Do we really want to show Mark Warner in a close race and then he wins by ten like everyone else had been saying?

And, of course, it did end up close, right? 
Oh, yeah. It was very close. But it’s the same kind of public-pressure thing. If someone were to say, “I think Trump’s going to win,” and go the full Trafalgar —

You never want to go full Trafalgar.
You don’t. But in Trafalgar’s defense, they were right and they got mocked for it. They got mocked for being basically right in 2016 and 2020.

But they also get mocked for being very opaque with their methods.
Yeah, there’s more to it than just that skew. But if your final answer this time is that Trump wins the popular vote by five and he loses it by one, you’re going to take a hit. No one wants to be that pollster. And so that’s kind of what happened, I think, in 2014, was they had these polls showing a close race, but no one else was showing it, and they didn’t want to be the odd pollster out. We know herding does occur to some degree, and it’s a legit concern.

This interview has been edited for length and clarity.